Genomic O Island 122, Locus for Enterocyte Effacement, and the Evolution of Virulent

JOURNAL OF BACTERIOLOGY, Sept. 2008, p. 5832–5840 Vol. 190, No. 170021-9193/08/$08.00�0 doi:10.1128/JB.00480-08Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Genomic O Island 122, Locus for Enterocyte Effacement, and theEvolution of Virulent Verocytotoxin-Producing Escherichia coli�†

Paulina Konczy,1 Kim Ziebell,1 Mariola Mascarenhas,1 Aileen Choi,1 Corinne Michaud,1Andrew M. Kropinski,1 Thomas S. Whittam,2 Mark Wickham,3

Brett Finlay,3 and Mohamed A. Karmali1*Laboratory for Foodborne Zoonoses, Public Health Agency of Canada, Guelph, Ontario N1G 3W4, Canada1; Microbial Evolution Laboratory,

National Food Safety and Toxicology Center, Michigan State University, East Lansing, Michigan 488242; and The University ofBritish Columbia, Michael Smith Laboratories, 301-2185 East Mall, Vancouver, British Columbia V6T 1Z4, Canada3

Received 8 April 2008/Accepted 17 June 2008

The locus of enterocyte effacement (LEE) and genomic O island 122 (OI-122) are pathogenicity islands inverocytotoxin-producing Escherichia coli (VTEC) serotypes that are associated with outbreaks and seriousdisease. Composed of three modules, OI-122 may occur as “complete” (with all three modules) or “incomplete”(with one or two modules) in different strains. OI-122 encodes two non-LEE effector (Nle) molecules that aresecreted by the LEE type III secretion system, and LEE and OI-122 are cointegrated in some VTEC strains.Thus, they are functionally linked, but little is known about the patterns of acquisition of these codependentislands. To examine this, we conducted a population genetics analysis, using multilocus sequence typing(MLST), with 72 VTEC strains (classified into seropathotypes A to E) and superimposed on the results theLEE and OI-122 contents of these organisms. The wide distribution of LEE and OI-122 modules among MLSTclonal groups corroborates the hypothesis that there has been lateral transfer of both pathogenicity islands.Sequence analysis of a pagC-like gene in OI-122 module 1 also revealed two nonsynonymous single-nucleotidepolymorphisms that could help discriminate a subset of seropathotype C strains and determine the presenceof the LEE. A nonsense mutation was found in this gene in five less virulent strains, consistent with a decayingor inactive gene. The modular nature of OI-122 could be explained by the acquisition of modules by lateraltransfer, either singly or as a group, and by degeneration of genes within modules. Correlations between clonalgroup, seropathotype, and LEE and OI-122 content provide insight into the role of genomic islands in VTECevolution.

Verocytotoxin-producing Escherichia coli (VTEC) and Shigatoxin-producing E. coli (STEC) are emerging zoonotic patho-gens consisting of multiple serotypes, over 200 of which havebeen isolated from cases of human disease (48). Human infec-tion by some VTEC serotypes, notably O157:H7, is associatedwith significant outbreaks and may lead to serious complica-tions, such as hemorrhagic colitis and the hemolytic-uremicsyndrome (HUS) (23, 25, 34). Karmali et al. (24) have classi-fied VTEC into five seropathotype groups based on the relativefrequency with which the serotypes are associated with seriousand epidemic human disease (Table 1). There is a strong as-sociation between VTEC seropathotypes A and B that causeserious or epidemic disease and the presence of two genomicislands, the locus of enterocyte effacement (LEE), which isassociated with the characteristic “attaching and effacing” le-sions (20, 34), and genomic O island 122 (OI-122) (24).

All of the virulence factors necessary for the formation ofthe attaching and effacing lesions in VTEC are encoded by theLEE pathogenicity island (18, 20, 34, 47), which encodes thestructural, accessory, and effector molecules of this type III

secretion system (4, 12, 13, 18, 19). The LEE of VTEC sero-type O157:H7 reference strain EDL 933, which is �43.4 kblong, contains 41 open reading frames which are organized infive polycistronic operons (LEE 1, LEE 2, LEE 3, LEE 5, andLEE 4) (34). Of particular interest in this study was LEE 5,which contains the eae gene, which encodes the outer mem-brane adhesin intimin (34). This operon also contains genesthat encode the translocated intimin receptor known as Tir(27) or EspE (7) and the Tir chaperone, CesT (1, 10).

OI-122 is a 23-kb pathogenicity island in O157:H7 strainEDL 933 which consists of three distinct modules separated bymobile genetic elements (Fig. 1) (24, 38). The first moduleencodes Z4321, a gene product with 46% homology to thephoP-activated gene C product (PagC) that enables survival inmacrophages of Salmonella enterica serovar Typhimurium (30,31, 36) (Fig. 1). This module is present in strains ranging fromstrains carrying a complete OI-122 with all three modules toincomplete strains carrying only this module. Module 2 carriesthe Z4326 (sen) gene, whose product is 39% homologous toShigella enterotoxin, and genes encoding two proteins, Z4328and Z4329, with 89 and 86% sequence homology to non-LEE-encoded (Nle) effectors of Citrobacter rodentium, NleB andNleE, respectively (8, 24). The third module encodes Z4332and Z4333, which are enterohemorrhagic E. coli (EHEC) fac-tors for adherence (Efa1 and Efa2).

OI-122 and LEE are functionally related; the LEE-encodedtype III secretory apparatus is required for secretion of the

* Corresponding author. Mailing address: Laboratory for Food-borne Zoonoses, Public Health Agency of Canada, 110 Stone Rd.West, Guelph, Ontario, Canada N1G 3W4. Phone: (519) 822-3300.Fax: (519) 822-2280. E-mail: [email protected].

† Supplemental material for this article may be found at http://jb.asm.org/.

� Published ahead of print on 27 June 2008.

5832

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

OI-122 non-LEE-encoded effectors NleB and NleE encoded inmodule 2 (8, 26). Multiple strains of a given serotype haveconserved patterns in the modular arrangement of OI-122genes (24). It has been proposed that the transposon-like in-dependent mobile elements of OI-122 are acquired or lost in amodular manner (46). The evidence which supports this pro-posal includes integration of OI-122 and LEE into each otherin various forms. While LEE is 0.7 Mb downstream of a com-plete OI-122 in EDL 933 (33, 38), mosaics of these two islandshave been identified by several groups. Shen et al. identified amosaic island in which OI-122 (with only module 1) was em-bedded in OI-48 in seropathotype C O113:H21 strain CL3(41). In the VTEC O26:H11-related RDEC strains, LEE isimmediately upstream of OI-122 (modules 2 and 3), cointe-grated as a 58-kb mosaic island (33, 44). A similar structurewas found in EHEC serotype O103:H2 strains, where LEE is43 kb downstream of OI-122 (modules 2 and 3), again physi-cally linked in this 111-kb hybrid island (22). Given that thepresence of both LEE and OI-122 is associated with the mostpathogenic VTEC strains, the objective of this study was toinvestigate the patterns of acquisition of these genomic islandsin VTEC. This was done by overlaying the distribution of LEEand OI-122 and seropathotypes on VTEC strain classificationby multilocus sequence typing (MLST).

MLST was chosen to shed light on the timelines for acqui-sition of these genomic islands in the evolutionary history of

this pathogen by using inferred relationships among the col-lection of seropathotype strains. In this study, the VTEC straingroup was analyzed using the MLST scheme developed at theSTEC Center (http://www.shigatox.net/stec/mlst-new/index.html). Using this method, phylogenetic relationships were in-ferred based on differences among “core” genomes deducedfrom an analysis of seven highly conserved housekeeping genes(see Table S1 in the supplemental material). These carefullyselected genes are presumably inherited vertically rather thanhorizontally and are subject to selective pressures, so there isslow, continual acquisition of random nucleotide changes (49).Studies have indicated that housekeeping genes diverge at arate that reflects the overall rate of genome divergence due tovertical and horizontal transfer events, as well as genome re-duction (29, 42, 49). The position of each node (strain) on atree based on MLST data inherently reflects this genome di-versity and provides a visual indication of how the genomesevolved relative to one another. The MLST technique wasused to generate clustering patterns for 72 VTEC strains,which were then analyzed to determine correlations betweenclonal groups, seropathotypes, and genomic island content.

MATERIALS AND METHODS

Bacterial strains. Seventy-two VTEC strains were used in this study (Fig. 2).The housekeeping gene sequences were obtained from the E. coli MLST(EcMLST) database for four other E. coli reference (ECOR) strains, includingthe O55:H7 progenitor strain, ECOR37, and for strain K-12, which was includedas a non-VTEC reference strain. Housekeeping gene sequences for the com-mensal organism E. coli HS (accession no. AAJY00000000), uropathogenic E.coli strain CFTO73 (accession no. AE014075), Shigella flexneri 2a (accession no.AE005674), and S. enterica serovar Typhimurium LT2 (accession no. AE006468)were retrieved from the NCBI GenBank database. These non-VTEC isolateswere added to give the MLST phylogenetic tree some additional structure and topermit us to draw further conclusions based on knowledge on their evolutionaryrelatedness to VTEC. Since it is known that S. enterica serovar Typhimuriumdiverged from E. coli �108 years ago (37), strain LT2 was selected for use as theoutgroup (root) in the MLST tree.

MLST. The housekeeping genes were PCR amplified from genomic DNAisolated from each of the 72 VTEC strains and K-12 using a Qiagen DNeasy kit(Qiagen Inc., Mississauga, Canada). The primers used for PCR amplification, aswell as sequencing, which were previously designed at the STEC Center (http://www.shigatox.net/stec/mlst-new/mlst_primers.html), are shown in Table S2 inthe supplemental material. The PCR amplification conditions were 35 cycles of92°C for 1 min, 58°C for 1 min, and 72°C for 30 s, with an initial denaturing stepof 94°C for 10 min for aspC, clpX, fadD, and lysP. For icdA, mdh, and uidA, ashorter extension time (15 s) at 72°C for 40 cycles was used. AmpliTaq Gold withbuffer II was used (Applied Biosystems, Foster City, CA) for increased specificity.

The amplicons were purified using a Qiagen PCR purification kit (Qiagen Inc.,

FIG. 1. Modular components of OI-122. ISA, insertion sequence-associated elements (or putative transposases) between the three modules.The PCR gene markers used to detect the presence of modules are indicated by bold type and blue.

TABLE 1. Classification of VTEC serotypes into fiveseropathotype groupsa

Seropathotype Relativeincidence Outbreaks Severe

disease Serotypes

A High Common Yes O157:H7, O157:NMB Moderate Uncommon Yes O26:H11, O103:H2,

O111:NM,O111:H8,O145:NM

C Low Rare Yes O91:H21,O104:H21,O113:H21, andothers

D Low Rare No MultipleE Nonhuman

onlyNAb NA Multiple

a See reference 24.b NA, not applicable.

VOL. 190, 2008 MOLECULAR EVOLUTION OF OI-122 5833

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

5834

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

Mississauga, Canada) and were sequenced using a DYEnamic ET terminatorcycle sequencing kit and a MegaBACE 500 automated DNA sequencer (Amer-sham Biosciences UK Ltd., Buckinghamshire, England). Sequencing of amplifiedfragments was done in both directions and in duplicate, so that for each gene aconsensus sequence was derived from four sequence reads using DiscoveryStudio Gene software (Accelrys Software Inc., San Diego, CA). The gene con-sensus sequences were aligned using ClustalX (45). For each strain, the seven-gene consensus sequence was concatenated using Molecular EvolutionaryGenetics Analysis (MEGA), version 3.1 (28), in the order aspC-clpX-fadD-icdA-lysP-mdh-uidA, giving an MLST “supergene” sequence that was 3,558 bp long(see Table S2 in the supplemental material).

Sequencing of the Z4321 (pagC-like) gene. The Z4321 locus was PCR ampli-fied using forward primer 5�-ATGAGTGGTTCAAGACTGG-3� and reverseprimer 5�-CCAACTCCAACAGTAAATCC-3�, yielding a 521-bp amplicon (24).This amplicon was sequenced as described above, and the sequence data werecropped to 399 bp prior to ClustalW alignment with Discovery Studio Genesoftware.

Data analysis. The “supergene” sequence was used to generate a dendrogramusing MEGA software (neighbor-joining algorithm with the Tajima-Nei model ofgenetic distance and bootstrapping of 1,000 replicates). The tree was subse-quently examined for patterns in the genomic island contents of strains in thecontext of their seropathotype and clonal group designations.

Correlations were also made between the presence of complete and incom-plete OI-122 and LEE in the context of the strains’ propensity to cause disease.Genes spanning all three modules of OI-122 (Z4321, Z4326, Z4332, and Z4333)(Fig. 1) and the eae locus of the LEE (Z5110) were previously amplified for theseropathotype collection (24). The primers used for PCR amplification weredescribed by Karmali et al. (24).

A protein tree of Z4321 was also made using MEGA (unweighted-pair groupmethod using average linkages algorithm with bootstrapping of 1,000 replicates)to determine subgroups of strains based on amino acid differences. The Nei-Gojobori procedure (28) was also performed using MEGA to evaluate thesubstitution rates.

RESULTS

Seropathotype, clonal group, and genomic island distribu-tions in the MLST tree. Among the VTEC isolates, seropatho-type A strains cluster as a distinct MLST clonal group carryingboth LEE and a complete OI-122 (Fig. 2). This O157 groupcorresponds to MLST clonal group 11 (EHEC 1) and is closelyrelated to the putative ancestral E. coli serogroup O55:H7strain (2, 3, 11). In the non-O157 branch, the LEE content andthe OI-122 forms vary within clonal groups. Seropathotype Bstrains are dispersed in four different MLST clusters, includingEHEC 2, EHEC-O121, and STEC 2. The fourth cluster, whichis less closely related to the other three clusters, all of whichshare a more recent common ancestor, consists of O145:NMstrains. Seropathotype B strains are positive for LEE and pos-sess either a complete or incomplete OI-122 (Fig. 2). Mostseropathotype C, D, and E strains occur in a wide diversity ofclonal groups, and these strains include members of MLSTgroups STEC 1, STEC 2, and EHEC-O121. Interestingly, onlymodule 1 is present when LEE is absent (Fig. 2). This may bedue to the absence of type III effectors (LEE associated) in thismodule. In contrast, module 2, which encodes effectors se-creted by LEE (NleB and NleE) correlates highly with LEE.Multiple strains of a serotype usually have the same pattern of

gene deletions or insertions in OI-122; for example, incom-plete OI-122 patterns are conserved in four strains belongingto serotype O113:H21 (LEE negative) and three strains be-longing to serotype O26:H11 (LEE positive).

S. enterica serovar Typhimurium LT2 differed significantly atthe genetic level from the VTEC isolates and other strains andso constitutes the outgroup node or root of the tree. S. flexneristrain 2a clustered together with uropathogenic E. coli strainCFT073 and two seropathotype D serotype O117:H7 strains.Reference strain K-12 clustered most closely with the othercommensal strain, strain HS, and with low-virulence strains(seropathotype D and E strains and environmental isolatesECOR-01 and ECOR-04).

Sequencing of the pagC-like Z4321 gene. Overall, 11 single-nucleotide polymorphisms (SNPs) and one indel (insertion ofadenine nucleotide) were found in the pagC-like gene of 43VTEC strains positive for this locus (see Table S3 in the sup-plemental material). The results for three O145:NM strainsand one O119:H25 strain had discrepancies with previous re-sults for the pagC-like locus (24) due to the presence of weakPCR bands. Of greatest interest in the SNP analysis were themutations that led to amino acid substitutions in the encodedprotein. Figure 3 shows the interrelatedness of Z4321-postivestrains and their groupings based on differences in proteinsequence. Sequence analysis of the pagC-like Z4321 locus re-vealed a nonsense mutation in five strains, three seropathotypeD strains (human serotype O103:H25 and O119:H25 strains)and two seropathotype E strains (bovine O98:H25 andO84:NM strains), as shown in Fig. 2. Strains with this decayedOI-122 module belong to a single MLST clonal group, group20, with high tree branch reliability (100% bootstrap support).They are grouped together despite variations in host (human andbovine), serotype (somatic antigens O103, O119, and O98 withflagellar antigen H25 and O84:NM), seropathotype (seropatho-types D and E), and genomic island content (LEE positive withincomplete and complete forms of OI-122). The mutation inthe Z4321 gene is the result of insertion of an A at nucleotide388, which led to a shift in the reading frame and changed thelast two codons of the protein product before the prematurestop (Fig. 3). Thus, this outer membrane protein is truncated atthe end of the third transmembrane loop and lacks the fourthand final loop of PagC. The nucleotide sequence downstreamof the stop codon is conserved among the five strains with thismutation. Among the five strains with this truncated geneproduct are the only human seropathotype D strains in theVTEC collection in which both LEE and OI-122 module 1 arepresent. These less virulent (seropathotype D) human isolateshave modules 1 and 2 but lacked module 3 (serotype O119:H25) or have a complete OI-122 (serotype O103:H25).

Furthermore, strains positive for the pagC-like locus fall intotwo groups, which are distinguishable by the presence of anonsynonymous SNP at nucleotide 207 of this gene (Fig. 3).

FIG. 2. Overlay of genomic island, clonal group, and seropathotype distributions relative to the MLST-inferred phylogenies (based on sevenhousekeeping genes) for 72 VTEC isolates. The box indicates the strains with a premature stop codon in a pagC-like gene, Z4321. The symbolsindicate the presence of a complete OI-122 and LEE (red triangles), the presence of an incomplete OI-122 and LEE (filled circles), the presenceof only OI-122 module 1 and the absence of LEE (open circles), and the absence of OI-122 and LEE (gray inverted triangles). n/a, gene was notanalyzed by PCR; CG, clonal group; ST, sequence type; UPEC, uropathogenic E. coli; EPEC, enteropathogenic E. coli; ShEt, Shigella enterotoxin.

VOL. 190, 2008 MOLECULAR EVOLUTION OF OI-122 5835

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

One group consists of eae (LEE)-positive strains that have Hisat the corresponding position in the Z4321 protein. The LEE-positive strains that have pagC have an incomplete OI-122(modules 1 and 2) or a complete OI-122 (all three modules).All of these strains (human seropathotypes A, B, and C) haveallelic variant 1 of pagC. The second group of Z4321-positivestrains (human seropathotype C and bovine seropathotypes Dand E) lack the LEE, have a Gln codon at bp 207, and do notcarry module 2 or 3. These LEE-negative strains exclusivelyharboring pagC have four SNPs (allelic variant 2) and sevenSNPs (allelic variant 3), respectively, compared to the variant1 allele (see Table S3 in the supplemental material). Note thatthe LEE-positive branch with the premature stop codon in thepagC-like locus differs only by a single adenine insertion atnucleotide 388 (human seropathotype D and bovine sero-pathotype E) from allelic variant 1 and comprises allelic vari-ant 4.

There is a division among seropathotype C strains withflagellar H21 antigen (O91:H21 and 0104:H21 versus O113:H21) as a result of a mutation at nucleotide 119 (Fig. 3).O113:H21 strains encode Tyr (TCT) at this location, and thisamino acid is unique to this group compared with all the otherZ4321-positive VTEC strains. The O113:H21 strains make upSTEC 2 clonal group 30, while the O91:H21 and O104:H21strains, which encode Ser (TAT), constitute STEC 1 clonalgroups 34 and 18, respectively.

G�C content. It was observed that the average G�C con-tent of the housekeeping genes (range, 51.5 to 54.0% [seeTable S2 in the supplemental material]) corresponds well withthe overall host genome base composition (E. coli K-12 [ac-cession no. NC_000913], 50.8%; EDL 933 [accession no.NC_002655], 50.4%). The average G�C content of the pagC-like Z4321 gene is 40.0%, which is low compared to that of theoverall genome, as expected for a gene on a pathogenicityisland (17).

Substitution rates. The Nei-Gojobori procedure (28) wasperformed using MEGA to evaluate the substitution rates forindividual housekeeping genes and the concatenated super-gene, as well as the pagC-like gene. More specifically, thenumbers of synonymous (pS) or silent and nonsynonymous(pN) substitutions leading to differences in amino acid se-quence per site were estimated for the housekeeping genes andthe pagC-like Z4321 locus. The assumption is that the rates ofevolution for a site are expected to be equal for neutral selec-tion (pS/pN � 1), whereas positive (diversifying) selection oc-curs when pN � pS and negative (purifying) selection occurswhen pS � pN (35). For each of the housekeeping genes andthe concatenated supergene, the rate of synonymous mutationis higher than the rate of nonsynonymous mutation (pS� pN),which implies that there is purifying selection (Table 2; seeTable S4 in the supplemental material), as expected. In thehousekeeping genes, the rate of synonymous mutation is ap-

100

85

92

85

100

LEEStrain Name Serotype Seropathotype/

DescriptionHost Module 1 Module 2

pagC -like Z4321

sen Z4326

efa1 Z4332

efa2 Z4333

eae Z5110

EDL933 O157:H7 A Control 1 + + + + +Sakai O157:H7 A Control 1 + + + + +D103F5 O157:H7 A Human 1 + + + + +278F1 (EC930550) O157:H7 A Human 1 + + + + +237F1 O157:H7 A Human 1 + + + + +E48F9 O157:H7 A Human 1 + + + + +157F1 O157:H7 A Human 1 + + + + +93111 (twec20010067) O157:H7 A Human 1 + + + + +OK1 (twec20010068) O157:H7 A Human 1 + + + + +158F2 O157:NM A Human 1 + + + + +279F1 (EC930549) O157:H7 A Human 1 + + + + +254 O157:H7 A Human 1 + + + + +E32511 (EC940098) O157:NM A Human 1 + + + + +ER6394 O157:NM A Human 1 + + + + +235F1 O157:H7 A Human 1 + + + + +Z3F3 O121:H19 B Human 1 + + + + +CL106 (EC930524) O121:H19 B Human 1 + + + + +274F4 O121:H19 B Human 1 + + + + +C69F1 O111:NM B Human 1 + + + + +CL101 O111:NM B Human 1 + + + + +R82F2 (EC930541) O111:NM B Human 1 + + + + +N994390 O121:NM C Human 1 + + + + +N994389 O121:NM C Human 1 + + + + +N004541 O5:NM C Human 1 + + + + +N004067 O5:NM C Human 1 + + + + +CL3 (EC930538) O113:H21 C Human 2 119, 207 Ser -> Tyr, His -> Gln + − − − −N993504 O113:H21 C Human 2 119, 207 Ser -> Tyr, His -> Gln + − − − −N900657 O113:H21 C Human 2 119, 207 Ser -> Tyr, His -> Gln + − − − −N890541 O113:H21 C Human 2 119, 207 Ser -> Tyr, His -> Gln + − − − −G5506 (twec20010077) O104:H21 C Human 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −B2F1 (EC930543) O91:H21 C Human 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −EC6990 O91:H21 C Human 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −EC6936 O91:H21 C Human 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −EC7181 O91:H21 C Human 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −EC920032 O171:H2 D Bovine 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −EC930480 O7:H4 D Bovine 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −EC920020 O156:NM E Bovine 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −EC940453 O88:H25 E Bovine 3 207, 443, 451 His -> Gln, Ser -> Phe, Ile -> Val + − − − −N022616 O103:H25 D Human 4 ins (388) Ser-Thr-Asp -> Ile-His-Stop + + + + +N004859 O103:H25 D Human 4 ins (388) Ser-Thr-Asp -> Ile-His-Stop + + + + +EC920267 O119:H25 D Human 4 ins (388) Ser-Thr-Asp -> Ile-His-Stop + + − − +EC930377 O98:H25 E Bovine 4 ins (388) Ser-Thr-Asp -> Ile-His-Stop + + − − +EC920044 O84:NM E Bovine 4 ins (388) Ser-Thr-Asp -> Ile-His-Stop + + − − +LT2 S. Typhimurium Control TNTD n/a2 n/a n/a n/a n/a

Allelic Variant

of Z4321

OI-122Module 3SNP location

in Z4321 (bp)1Amino acid change caused by SNP,

relative to EDL933 Z4321 control

(1) SNP location is relative to published EDL 933 Z4321 gene (NCBI Accession # NC_002655).(2) Z4321 has known homology to PagC in Salmonella typhimurium LT2 (NCBI Accession # AE006468).

FIG. 3. Phylogenetic tree based on PagC-like Z4321 protein in 43 VTEC strains. SNPs in Z4321 in S. enterica serovar Typhimurium were toonumerous to display (TNTD). For an explanation of the symbols, see the legend to Fig. 2. n/a, gene was not analyzed.

5836 KONCZY ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

proximately 42-fold higher than the rate of nonsynonymousmutation (Table 2). The most divergent housekeeping gene(i.e., the least conserved gene) with the highest rate of non-synonymous substitution is uidA (see Table S4 in the supple-mental material). The rate of nonsynonymous substitution was20-fold higher in pagC than in the housekeeping genes, whichis consistent with a divergent gene; further, the rate of synon-ymous substitution was 1.6 times lower.

The rates of substitution were also analyzed separately foreach of the five seropathotypes (seropathotypes A through E)to test the hypothesis that the individual groups evolve withdifferent substitution rates. There were general increases in therates of both synonymous substitution and nonsynonymoussubstitution in the less virulent seropathotypes in both thehousekeeping gene (supergene) and pagC-like gene sequences(Table 2). In seropathotype D and E strains there was a sig-nificantly higher rate of mutation (especially pN) in pagC thanin the supergene (P � 0.05, Student-Newman-Keuls posttest).The pS/pN value is closer to 1 for pagC in these strains, con-sistent with a shift toward neutral selection after the introduc-tion of the stop codon (gene inactivation).

DISCUSSION

The predicted substitution rates and the G�C content of thehousekeeping genes compared to that of the overall genomeconfirm that these genes are being evolutionarily conserved inVTEC genomes, thereby validating the use of MLST for de-fining evolutionarily related clonal groups. These observationsconfirm our assumption that the selected housekeeping genesare vertically acquired and so provide a snapshot of the diver-gence that occurs in the overall genome (49).

E. coli O157:H7 strains associated with outbreaks and severeepidemicity have been shown to represent a single phyloge-netic branch when they are grouped by MLST (using sevenhousekeeping genes), comprising 100% of the seropathotypeA strains (40, 50, 51). It has been postulated that over the last50 years, the pathogenic O157 lineage has evolved from anenteropathogenic E. coli O55:H7 group with the acquisition ofverotoxin-converting phages and an O157 rfb (O-antigen sub-unit) gene cluster (3). This finding was corroborated in the

current study, where all of the O157 strains clustered as asingle group whose nearest neighbor was the enteropathogenicE. coli O55 strain (Fig. 2). The O157 group of strains and theirO55 ancestor have either converged or diverged from non-O157 VTEC at some point in their evolutionary history. Thisseparation may coincide with the evolutionary split of O157and K-12, which occurred 4.5 million years ago (21). S. flexneri2a, which mapped among the outliers from the major non-O157 cluster, is more closely related to K-12 than to EDL 933(O157 seropathotype A) (21), and this was confirmed by thefinding that these organisms share a more recent commonancestral node in the tree. The facts that uidA was found to bethe least conserved of the housekeeping genes and that it wasabsent only in S. enterica serovar Typhimurium LT2, which isthe outlier strain in the MLST tree, reaffirm the structure ofthe tree. There have been more evolutionary splits from Sal-monella in the non-O157 clusters than in the O157 cluster.

The highly branching non-O157 group reflects a high degreeof genetic rearrangement compared to the O157 cluster. It canbe postulated that losing genetic factors and moving fromvirulent to less virulent may give new non-O157 variants aselective advantage in surviving and/or in contributing topathogenesis during VTEC infection. Some of the internalnon-O157 branches in the tree may represent the fastest-evolv-ing strains (including seropathotype D and E strains with anonsense mutation in pagC-like gene Z4321), and a lot ofvariation in genomic island content has been observed withinthese closely related subclusters. High substitution rates innon-O157 (seropathotype C, D, and E) strains corroborate thisobservation. Examination of the occurrence of the LEE andcomponents of OI-122 in widely divergent MLST clonal groupshas provided striking evidence of horizontal transfer of chro-mosomal genes and pathogenicity islands. Furthermore, thisstudy, using OI-122 as an example, provides novel insights intothe acquisition and fate of island components.

Evidence of horizontal gene transfer among non-O157strains. The O157 EHEC 1 group represents the only VTECgroup for which there is a direct correlation between sero-pathotype (seropathotype A) and genomic island content(LEE positive with complete OI-122). Otherwise, among the

TABLE 2. Rates of synonymous and nonsynonymous substitution in Z4321 and in the MLST supergene sequence based on Nei-Gojobori andJukes-Cantor analysis for the overall VTEC collection and for seropathotypes

Gene Sero-pathotype Length(bp) pS (102) SE for pS (102)a pN (102) SE for pN (102)a Conclusion Significance

pagC-like Z4321 A (n � 15) 399 0.00 0.0000 0.00 0.0000 None NoneB (n � 6) 399 0.00 0.0000 0.00 0.0000 None NoneC (n � 13) 399 2.26 0.0107 0.65 0.0031 pS � pN Purifying selectionD (n � 5) 399 8.52 0.0265 7.47 0.0170 pS � pN Purifying selection (NS)b

E (n � 4) 399 9.47 0.0287 8.30 0.0198 pS � pN Purifying selection (NS)Total (n � 43) 399 3.69 0.0117 2.83 0.0067 pS � pN Purifying selection (NS)

MLST supergene A (n � 15) 3,558 0.00 0.0000 0.00 0.0000 None NoneB (n � 15) 3,558 3.21 0.0039 0.06 0.0003 pS � pN Purifying selectionC (n � 14) 3,558 3.23 0.0043 0.07 0.0003 pS � pN Purifying selectionD (n � 14) 3,558 5.65 0.0050 0.12 0.0004 pS � pN Purifying selectionE (n � 14) 3,558 3.99 0.0044 0.08 0.0003 pS � pN Purifying selectionTotal (n � 72) 3,558 5.82 0.0047 0.14 0.0004 pS � pN Purifying selection

a Based on the estimated substitution rates.b NS, nonsignificant difference between pS and pN (P � 0.05, Student-Newman-Keuls test), implying a shift toward neutral selection.

VOL. 190, 2008 MOLECULAR EVOLUTION OF OI-122 5837

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

non-O157 VTEC strains, it is clear from the MLST clusteringpatterns that seropathotype and genomic island distributionare not clonally restricted. In fact, seropathotypes are widelydispersed throughout the tree and are more widely dispersedwith decreasing level of epidemicity (seropathotype A clusterson one branch, seropathotype B clusters on four branches,seropathotype C clusters on five branches, seropathotype Dclusters on nine branches, and seropathotype E clusters onnine branches [Fig. 2]). LEE and the various OI-122 forms(complete, incomplete, or absent) are widely distributed indifferent lineages (clonal groups) and seropathotypes. Modularcomponents of OI-122 that are variably present or absent on abranch containing closely related strains likely were obtainedfrom less closely related strains or species via horizontal genetransfer. For example, examination of the MLST tree aroundseropathotype C, D, and E strains shows that there is a mixtureof LEE-positive and -negative strains in a branch. This scat-tered distribution of eae genes in the MLST tree is character-istic of a horizontal transfer event when a set of genes has beenintroduced into a lineage (9). Among seropathotype B throughE strains, OI-122 molecules and their gene content are simi-larly scattered throughout the tree. For example, in brancheswhere all strains belong to the same serotype and have thesame incomplete form of OI-122, it is likely that componentswere acquired in a modular manner and became stabilized inthe genome. The observations for LEE and OI-122 describedabove support the notion that the virulence genes comprisingthese gene cassettes have been and continue to be horizontallytransferred across lineages. The wide MLST clonal distributionof these two islands and the lack of association between sero-pathotype, genomic island content, OI-122 module content,and MLST clustering patterns are also indicative of horizontaltransfer among strains (Fig. 2).

pagC-like Z4321 deletion mutants in less virulent strains.The evidence of decay of OI-122 elements in module 1(Z4321) in five strains that belong to seropathotypes D and Ecorrelates with the apparent reduced virulence of these LEE-positive strains. It also strongly indicates that there has beenhorizontal gene transfer since this mutated genetic element isshared by strains whose seropathotype and genomic islandprofiles and hosts differ but the strains are closely relatedphylogenetically (15). The sequence flanking the indel (an ad-enine insertion), particularly downstream of the prematurestop codon, which no longer encodes a functional proteinproduct, is conserved in these different strains. Based on theminimal assumption of evolution, this insertion was introducedonce at some point in the evolutionary history of this collectionand was passed horizontally among the strains (15). The OI-122 modular patterns may reflect horizontal acquisition of oneor more modules independently or modular decay followingtransfer of a complete OI-122 (Fig. 3). A correspondingly highrate of synonymous and nonsynonymous (detrimental) substi-tutions in the Z4321 gene in these less virulent seropathotypesis also consistent with a decaying or inactive gene. Functionalprotein studies with Yersinia and Salmonella involving closelyrelated Ail and Rck proteins indicated that the fourth extra-cellular loop (absent in the truncated Z4321 protein) is notassociated with adhesion, invasion, or serum resistance pheno-types (5, 32, 36). The third loop (full length in the truncatedZ4321 protein) has been shown to confer virulence properties

in Rck (5). In the future, functional assays may be performedto assess the impact of this mutation on the Z4321 protein inVTEC. Wickham et al. showed that there is a significant asso-ciation between the presence of a combination of pagC and sen(ent), nleB, and efa-1/lifA and HUS after infection in non-O157E. coli (46). On its own, the pagC-like gene is associated withHUS but not with outbreaks (46). It is interesting that theseropathotype E strains in this study which had the mutatedpagC gene were of bovine origin and also did not contain efa-1(Fig. 2). It has been proposed that the additive effect of thesetwo genes contributes significantly to causing HUS (6, 46).While the pagC locus may contribute to pathogenesis in morevirulent VTEC, pseudogenization may have hampered its ac-tivity and given rise to these less virulent variants. The obser-vation that human strains without this deleterious mutation inpagC (on a complete OI-122 with LEE present) are seropatho-type A, B, or C strains and strains with the truncated gene (ona complete or incomplete OI-122 with LEE) are seropathotypeD strains supports this theory.

SNPs in Z4321 useful for differentiation of O113:H21 andthe presence of LEE. Human seropathotype C strains withflagellar H21 antigen were originally classified in one clonalgroup (http://www.shigatox.net/cgi-bin/stec/clonal). Data fromthis study show that these strains belong to different clonalgroups (Fig. 2), and this was corroborated by a SNP in Z4321(A 3 C at bp 119) that results in an O113:H21-specific Tyrresidue (in clonal group 30) instead of Ser, which is found in allthe other serotypes tested, including O91:H21 (clonal group34) and O104:H21 (clonal group 18). While they share a com-mon flagellar H21 antigen and have the same OI-122 and LEEprofiles (LEE negative and OI-122 module 1 only [Fig. 2]),these groups split at some point in their evolutionary history.There may be other genomic differences between these groups,but targeting this SNP may be a quick way to differentiatebetween the H21 clonal groups and to screen for the O113:H21serotype.

A second nonsynonymous SNP at nucleotide 207 of theZ4321 gene allowed us to predict the presence of LEE becausestrains having His at the corresponding position harbor eae,while strains with Gln lack eae. An interesting observation isthat only the strains that carry the pagC-like gene exclusively(modules 2 and 3 are not present) both have this nonsynony-mous substitution (His3 Gln) and lack LEE. Strains with thisproperty include O113:H21 strain CL3, in which Z4321 is partof a mosaic island cointegrated with OI-48 (41). Screening fora marker, Z1640::S1, that is indicative of this hybrid islandindicated that other serotypes in the VTEC collection with thischaracteristic include O156:NM, O171:H2, O7:H4, O88:H25,and O91:H21 (41). It is unclear whether the pagC-like genefirst appeared in strains such as O157:H7 strain EDL 933 aspart of a complete OI-122 or as part of the OI-122::OI-48mosaic island, as observed in O113:H21 strain CL3. The pagCalleles (alleles 1 and 4) of LEE-positive strains have four toeight nucleotide differences compared with the allelic variants(alleles 2 and 3) of the LEE-negative lineages (see Table S3 inthe supplemental material). The pagC-like alleles may havebeen exchanged between the LEE-positive and -negative lin-eages at some point, along with the acquisition of SNPs. Giventhat there are more than a few nucleotide differences between

5838 KONCZY ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

these genes, the possibility that the genes may also have arisenfrom a separate ancestor cannot be overlooked.

Concluding remarks. Bacterial evolution is driven by theneed to achieve optimal “fitness,” a concept that refers toattributes that enhance the survival, spread, and/or transmis-sion of an organism within a specific ecological niche (16, 39).Horizontal gene transfer and gene degradation provide mech-anisms for rapid adaptation to changing ecological circum-stances or for acquiring optimal fitness so that an organism cansurvive and flourish under such circumstances (16). The evo-lutionary advantage of acquiring genomic islands over acquir-ing smaller genetic elements is that a large number of genesencoding many complementary functions may be transferreden bloc to the recipient organism, a process that may result in“evolution in quantum leaps” (14); one example of this is theacquisition of a type III secretion system which is encoded byLEE (17). On the other hand, a minor environmental changemay not require acquisition of all the genetic material presentin a genomic island, and the transfer of smaller elements, suchas plasmids or transposons, may be more efficient. Consideringthat we did not explore the LEE in this study to the sameextent as OI-122, further analysis should shed more light onthe interplay of these genomic islands. OI-122 has three mod-ules, each consisting of genes associated with mobile geneticelements, including transposase genes. One or more of theseelements may thus be transposons, a concept supported by theoccurrence of one, two, or three OI-122 modules in individualstrains. Transposons are typically associated with the transferof antimicrobial resistance genes under the selective pressureof antibiotics. However, transposons containing genes that en-code catabolic functions have also been described, and theirpresence may be selected by specific substrates (43). Environ-mental selective factors that could be involved in selectingspecific OI-122 modules remain to be investigated. Knowledgeabout the ecological determinants of the presence, absence, ordecay of specific OI-122 modules could provide new insightsinto the origin of pathogenic clones expressing specific modu-lar patterns.

This population genetics study provided new insights abouttwo genomic islands in the evolution of pathogenic VTEC. Theresults support the hypothesis that genomic islands in VTECare horizontally acquired and that some of them, like OI-122,are likely acquired in a modular manner. It appears that theless virulent VTEC strains have experienced a loss of genomicisland components. Further work can address the question ofwhat role the horizontally acquired islands play in the emer-gence of new pathogens.

ACKNOWLEDGMENTS

We thank Shelley Frost for technical assistance.This research was supported by the Public Health Agency of Can-

ada, as well as by a Canadian Institutes of Health Research Food andWater Safety grant and by operating grants from the Canadian Insti-tutes of Health Research and the Howard Hughes Medical Institute.

REFERENCES

1. Abe, A., M. de Grado, R. A. Pfuetzner, C. Sanchez-Sanmartin, R. Devinney,J. L. Puente, N. C. Strynadka, and B. B. Finlay. 1999. EnteropathogenicEscherichia coli translocated intimin receptor, Tir, requires a specific chap-erone for stable secretion. Mol. Microbiol. 33:1162–1175.

2. Allen, N. L., A. C. Hilton, R. Betts, and C. W. Penn. 2001. Use of represen-tational difference analysis to identify Escherichia coli O157-specific DNAsequences. FEMS Microbiol. Lett. 197:195–201.

3. Brussow, H., C. Canchaya, and W. D. Hardt. 2004. Phages and the evolutionof bacterial pathogens: from genomic rearrangements to lysogenic conver-sion. Microbiol. Mol. Biol. Rev. 68:560–602.

4. Chan, K. N., A. D. Phillips, S. Knutton, H. R. Smith, and J. A. Walker-Smith.1994. Enteroaggregative Escherichia coli: another cause of acute and chronicdiarrhoea in England? J. Pediatr. Gastroenterol. Nutr. 18:87–91.

5. Cirillo, D. M., E. J. Heffernan, L. Wu, J. Harwood, J. Fierer, and D. G.Guiney. 1996. Identification of a domain in Rck, a product of the Salmonellatyphimurium virulence plasmid, required for both serum resistance and cellinvasion. Infect. Immun. 64:2019–2023.

6. Coombes, B. K., M. E. Wickham, M. Mascarenhas, S. Gruenheid, B. B.Finlay, and M. A. Karmali. 2008. Molecular analysis as an aid to assess thepublic health risk of non-O157 Shiga toxin-producing Escherichia coli. Appl.Environ. Microbiol. 74:2153–2160.

7. Deibel, C., S. Kramer, T. Chakraborty, and F. Ebel. 1998. EspE, a novelsecreted protein of attaching and effacing bacteria, is directly translocatedinto infected host cells, where it appears as a tyrosine-phosphorylated 90 kDaprotein. Mol. Microbiol. 28:463–474.

8. Deng, W., J. L. Puente, S. Gruenheid, Y. Li, B. A. Vallance, A. Vazquez, J.Barba, J. A. Ibarra, P. O’Donnell, P. Metalnikov, K. Ashman, S. Lee, D.Goode, T. Pawson, and B. B. Finlay. 2004. Dissecting virulence: systematicand functional analyses of a pathogenicity island. Proc. Natl. Acad. Sci. USA101:3597–3602.

9. Dutta, C., and A. Pan. 2002. Horizontal gene transfer and bacterial diversity.J. Biosci. 27:27–33.

10. Elliott, S. J., S. W. Hutcheson, M. S. Dubois, J. L. Mellies, L. A. Wainwright,M. Batchelor, G. Frankel, S. Knutton, and J. B. Kaper. 1999. Identificationof CesT, a chaperone for the type III secretion of Tir in enteropathogenicEscherichia coli. Mol. Microbiol. 33:1176–1189.

11. Feng, P., K. A. Lampel, H. Karch, and T. S. Whittam. 1998. Genotypic andphenotypic changes in the emergence of Escherichia coli O157:H7. J. Infect.Dis. 177:1750–1753.

12. Foubister, V., I. Rosenshine, M. S. Donnenberg, and B. B. Finlay. 1994. TheeaeB gene of enteropathogenic Escherichia coli is necessary for signal trans-duction in epithelial cells. Infect. Immun. 62:3038–3040.

13. Goosney, D. L., R. DeVinney, and B. B. Finlay. 2001. Recruitment of cy-toskeletal and signaling proteins to enteropathogenic and enterohemor-rhagic Escherichia coli pedestals. Infect. Immun. 69:3315–3322.

14. Groisman, E. A., and H. Ochman. 1996. Pathogenicity islands: bacterialevolution in quantum leaps. Cell 87:791–794.

15. Gupta, R. S. 1998. Protein phylogenies and signature sequences: a reap-praisal of evolutionary relationships among archaebacteria, eubacteria, andeukaryotes. Microbiol. Mol. Biol. Rev. 62:1435–1491.

16. Hacker, J., and E. Carniel. 2001. Ecological fitness, genomic islands andbacterial pathogenicity. A Darwinian view of the evolution of microbes.EMBO Rep. 2:376–381.

17. Hacker, J., and J. B. Kaper. 2000. Pathogenicity islands and the evolution ofmicrobes. Annu. Rev. Microbiol. 54:641–679.

18. Jarvis, K. G., J. A. Giron, A. E. Jerse, T. K. McDaniel, M. S. Donnenberg,and J. B. Kaper. 1995. Enteropathogenic Escherichia coli contains a putativetype III secretion system necessary for the export of proteins involved inattaching and effacing lesion formation. Proc. Natl. Acad. Sci. USA 92:7996–8000.

19. Jarvis, K. G., and J. B. Kaper. 1996. Secretion of extracellular proteins byenterohemorrhagic Escherichia coli via a putative type III secretion system.Infect. Immun. 64:4826–4829.

20. Jerse, A. E., J. Yu, B. D. Tall, and J. B. Kaper. 1990. A genetic locus ofenteropathogenic Escherichia coli necessary for the production of attachingand effacing lesions on tissue culture cells. Proc. Natl. Acad. Sci. USA87:7839–7843.

21. Jin, Q., Z. Yuan, J. Xu, Y. Wang, Y. Shen, W. Lu, J. Wang, H. Liu, J. Yang,F. Yang, X. Zhang, J. Zhang, G. Yang, H. Wu, D. Qu, J. Dong, L. Sun, Y.Xue, A. Zhao, Y. Gao, J. Zhu, B. Kan, K. Ding, S. Chen, H. Cheng, Z. Yao,B. He, R. Chen, D. Ma, B. Qiang, Y. Wen, Y. Hou, and J. Yu. 2002. Genomesequence of Shigella flexneri 2a: insights into pathogenicity through com-parison with genomes of Escherichia coli K12 and O157. Nucleic Acids Res.30:4432–4441.

22. Jores, J., S. Wagner, L. Rumer, J. Eichberg, C. Laturnus, P. Kirsch, P.Schierack, H. Tschape, and L. H. Wieler. 2005. Description of a 111-kbpathogenicity island (PAI) encoding various virulence features in the entero-hemorrhagic E. coli (EHEC) strain RW1374 (O103:H2) and detection of asimilar PAI in other EHEC strains of serotype 0103:H2. Int. J. Med. Micro-biol. 294:417–425.

23. Karmali, M. A. 1989. Infection by verocytotoxin-producing Escherichia coli.Clin. Microbiol. Rev. 2:15–38.

24. Karmali, M. A., M. Mascarenhas, S. Shen, K. Ziebell, S. Johnson, R. Reid-Smith, J. Isaac-Renton, C. Clark, K. Rahn, and J. B. Kaper. 2003. Associ-ation of genomic O island 122 of Escherichia coli EDL 933 with verocyto-toxin-producing Escherichia coli seropathotypes that are linked to epidemicand/or serious disease. J. Clin. Microbiol. 41:4930–4940.

25. Karmali, M. A., M. Petric, C. Lim, P. C. Fleming, G. S. Arbus, and H. Lior.1985. The association between idiopathic hemolytic uremic syndrome and

VOL. 190, 2008 MOLECULAR EVOLUTION OF OI-122 5839

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.

infection by verotoxin-producing Escherichia coli. J. Infect. Dis. 151:775–782.

26. Kelly, M., E. Hart, R. Mundy, O. Marches, S. Wiles, L. Badea, S. Luck, M.Tauschek, G. Frankel, R. M. Robins-Browne, and E. L. Hartland. 2006.Essential role of the type III secretion system effector NleB in colonizationof mice by Citrobacter rodentium. Infect. Immun. 74:2328–2337.

27. Kenny, B., R. DeVinney, M. Stein, D. J. Reinscheid, E. A. Frey, and B. B.Finlay. 1997. Enteropathogenic E. coli (EPEC) transfers its receptor forintimate adherence into mammalian cells. Cell 91:511–520.

28. Kumar, S., K. Tamura, and M. Nei. 2004. MEGA3: integrated software formolecular evolutionary genetics analysis and sequence alignment. Brief.Bioinform. 5:150–163.

29. Maiden, M. C., J. A. Bygraves, E. Feil, G. Morelli, J. E. Russell, R. Urwin,Q. Zhang, J. Zhou, K. Zurth, D. A. Caugant, I. M. Feavers, M. Achtman, andB. G. Spratt. 1998. Multilocus sequence typing: a portable approach to theidentification of clones within populations of pathogenic microorganisms.Proc. Natl. Acad. Sci. USA 95:3140–3145.

30. Miller, S. I. 1991. PhoP/PhoQ: macrophage-specific modulators of Salmo-nella virulence? Mol. Microbiol. 5:2073–2078.

31. Miller, S. I., W. S. Pulkkinen, M. E. Selsted, and J. J. Mekalanos. 1990.Characterization of defensin resistance phenotypes associated with muta-tions in the phoP virulence regulon of Salmonella typhimurium. Infect. Im-mun. 58:3706–3710.

32. Miller, V. L., K. B. Beer, G. Heusipp, B. M. Young, and M. R. Wachtel. 2001.Identification of regions of Ail required for the invasion and serum resis-tance phenotypes. Mol. Microbiol. 41:1053–1062

33. Morabito, S., R. Tozzoli, E. Oswald, and A. Caprioli. 2003. A mosaic patho-genicity island made up of the locus of enterocyte effacement and a patho-genicity island of Escherichia coli O157:H7 is frequently present in attachingand effacing E. coli. Infect. Immun. 71:3343–3348.

34. Nataro, J. P., and J. B. Kaper. 1998. Diarrheagenic Escherichia coli. Clin.Microbiol. Rev. 11:142–201.

35. Nei, M., and T. Gojobori. 1986. Simple methods for estimating the numbersof synonymous and nonsynonymous nucleotide substitutions. Mol. Biol.Evol. 3:418–426.

36. Nishio, M., N. Okada, T. Miki, T. Haneda, and H. Danbara. 2005. Identifi-cation of the outer-membrane protein PagC required for the serum resis-tance phenotype in Salmonella enterica serovar Choleraesuis. Microbiology151:863–873.

37. Ochman, H., and A. C. Wilson. 1987. Evolution in bacteria: evidence for auniversal substitution rate in cellular genomes. J. Mol. Evol. 26:74–86.

38. Perna, N. T., G. Plunkett III, V. Burland, B. Mau, J. D. Glasner, D. J. Rose,G. F. Mayhew, P. S. Evans, J. Gregor, H. A. Kirkpatrick, G. Posfai, J.Hackett, S. Klink, A. Boutin, Y. Shao, L. Miller, E. J. Grotbeck, N. W. Davis,A. Lim, E. T. Dimalanta, K. D. Potamousis, J. Apodaca, T. S. Ananthara-man, J. Lin, G. Yen, D. C. Schwartz, R. A. Welch, and F. R. Blattner. 2001.Genome sequence of enterohaemorrhagic Escherichia coli O157:H7. Nature409:529–533.

39. Preston, G. M., B. Haubold, and P. B. Rainey. 1998. Bacterial genomics andadaptation to life on plants: implications for the evolution of pathogenicityand symbiosis. Curr. Opin. Microbiol. 1:589–597.

40. Reid, S. D., C. J. Herbelin, A. C. Bumbaugh, R. K. Selander, and T. S.Whittam. 2000. Parallel evolution of virulence in pathogenic Escherichiacoli. Nature 406:64–67.

41. Shen, S., M. Mascarenhas, K. Rahn, J. B. Kaper, and M. A. Karmali. 2004.Evidence for a hybrid genomic island in verocytotoxin-producing Escherichiacoli CL3 (serotype O113:H21) containing segments of EDL933 (serotypeO157:H7) O islands 122 and 48. Infect. Immun. 72:1496–1503.

42. Stackebrandt, E., W. Frederiksen, G. M. Garrity, P. A. Grimont, P.Kampfer, M. C. Maiden, X. Nesme, R. Rossello-Mora, J. Swings, H. G.Truper, L. Vauterin, A. C. Ward, and W. B. Whitman. 2002. Report of theAd Hoc Committee for the Re-Evaluation of the Species Definition inBacteriology. Int. J. Syst. Evol. Microbiol. 52:1043–1047.

43. Tan, H. M. 1999. Bacterial catabolic transposons. Appl. Microbiol. Biotech-nol. 51:1–12.

44. Tauschek, M., R. A. Strugnell, and R. M. Robins-Browne. 2002. Character-ization and evidence of mobilization of the LEE pathogenicity island ofrabbit-specific strains of enteropathogenic Escherichia coli. Mol. Microbiol.44:1533–1550.

45. Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G.Higgins. 1997. The CLUSTAL_X Windows interface: flexible strategies formultiple sequence alignment aided by quality analysis tools. Nucleic AcidsRes. 25:4876–4882.

46. Wickham, M. E., C. Lupp, M. Mascarenhas, A. Vazquez, B. K. Coombes,N. F. Brown, B. A. Coburn, W. Deng, J. L. Puente, M. A. Karmali, and B. B.Finlay. 2006. Bacterial genetic determinants of non-O157 STEC outbreaksand hemolytic-uremic syndrome after infection. J. Infect. Dis. 194:819–827.

47. Wieler, L. H., T. K. McDaniel, T. S. Whittam, and J. B. Kaper. 1997.Insertion site of the locus of enterocyte effacement in enteropathogenic andenterohemorrhagic Escherichia coli differs in relation to the clonal phylog-eny of the strains. FEMS Microbiol. Lett. 156:49–53.

48. World Health Organization. 1999. Zoonotic non-O157 Shiga toxin-produc-ing Escherichia coli (STEC). Report of a WHO scientific working groupmeeting. World Health Organization, Geneva, Switzerland.

49. Zeigler, D. R. 2003. Gene sequences useful for predicting relatedness ofwhole genomes in bacteria. Int. J. Syst. Evol. Microbiol. 53:1893–1900.

50. Zhang, W., W. Qi, T. J. Albert, A. S. Motiwala, D. Alland, E. K. Hyytia-Trees,E. M. Ribot, P. I. Fields, T. S. Whittam, and B. Swaminathan. 2006. Probinggenomic diversity and evolution of Escherichia coli O157 by single nucleo-tide polymorphisms. Genome Res. 16:757–767.

51. Ziebell, K., P. Konczy, I. Yong, S. Frost, M. Mascarenhas, A. M. Kropinski,T. S. Whittam, S. C. Read, and M. A. Karmali. 2008. Applicability ofphylogenetic methods for characterizing the public health significance ofverocytotoxin-producing Escherichia coli strains. Appl. Environ. Microbiol.74:1671–1675.

5840 KONCZY ET AL. J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

23

Nov

embe

r 20

21 b

y 21

7.24

.169

.144

.